Monitoring Lake Ice Seasons in Southwest Alaska with Modis Images

MONITORING LAKE ICE SEASONS IN SOUTHWEST ALASKA WITH MODIS IMAGES

Page Spencer, Chief of Natural Resources Lake Clark National Park and Preserve 240 W 5th Ave Anchorage, Alaska 99501 [email protected]

Amy E. Miller, Ecologist National Park Service, Southwest Alaska Network 240 W 5th Ave Anchorage, Alaska 99501 [email protected]

Bradley Reed. Associate Program Coordinator U.S. Geological Survey (USGS) Reston, VA 20191 [email protected]

Michael Budde, Geographer USGS Center for Earth Resources Observation and Science (EROS) Sioux Falls, SD 57198 [email protected]

ABSTRACT

Impacts of global climate change are expected to result in greater variation in the seasonality of snowpack, lake ice, and vegetation dynamics in southwest Alaska. All have wide-reaching physical and biological ecosystem effects in the region. We used Moderate Resolution Imaging Spectroradiometer (MODIS) Rapidfire browse images, snow cover extent, and vegetation index products for interpreting interannual variation in the duration and extent of snowpack, lake ice, and vegetation dynamics for southwest Alaska. The approach integrates multiple seasonal metrics across large ecological regions. The general annual pattern of the lake ice season in southwest Alaska begins as water cools when fall storms sweep in from the North Pacific. Ice forms on high altitude or small lakes by Oct/Nov, with larger lakes usually freezing over by Dec/Jan. Warm mid-winter storms often hit the study area, bringing wind, rain and melting temperatures that result in several inches of water on the ice or partial break-up of the ice pack. The return of cold, still Arctic high pressure refreezes the lakes until spring break-up in April and May. Throughout the observation period (2001–2007), snow cover duration was stable within ecoregions, with variable start and end dates. The start of the lake ice season lagged the snow season by 2 to 3 months. Within a given lake, freeze-up dates varied in timing and duration, while break-up dates were more consistent. Vegetation phenology varied less than snow and ice metrics, with start-of-season dates comparatively consistent across years. Higher than average temperatures during the El Niño winter of 2002-2003 were expressed in anomalous ice and snow season patterns. We are developing a consistent, MODIS-based dataset that will be used to monitor temporal trends of each of these seasonal metrics and to map areas of change for the study area.

INTRODUCTION

Ecosystem responses to changing climate are expected to occur over different temporal scales, with the timing of seasonal events (e.g., freeze-thaw cycles, snowpack formation and snowmelt, and the onset or end of the growing season) showing a great deal of variability. Long-term changes in snowpack, lake ice, and vegetation may be particularly pronounced in high northern latitudes where small changes in climate can alter the timing of these events. Remotely sensed data can potentially capture much of this variation in seasonality, and provided that the

Pecora 17 – The Future of Land Imaging…Going Operational November 18 – 20, 2008 Denver, Colorado data record is sufficiently long, enable us to distinguish between interannual variability and multi-year trends (e.g., in snowpack duration or growing season length). We describe the use of Moderate Resolution Imaging Spectroradiometer (MODIS) data from Aqua and Terra satellites to monitor lake ice seasonal dynamics in southwest Alaska. This paper focuses on the lake ice metrics, including the timing and duration of ice formation and break-up on large lakes as manually interpreted from Rapidfire MODIS images. The moderate spatial resolution (250–1000 m) and high temporal resolution provided by MODIS enables frequent observations of a region characterized by a rapidly changing landscape, complex climatic regime, frequent cloud cover, and little infrastructure to support ground-based measurements. The National Park Service (NPS) is also monitoring the landscape dynamics of snow and growing seasons for the study area using standard MODIS snow and NDVI products. Summary metrics for snow cover and vegetation phenology (start and end of season) were also derived using standard MODIS products for snow and NDVI (Reed et al, in press). We examined variation in each metric across years (2001-2007) and the relationships among metrics.

Background Increasing temperatures projected for the northern high latitudes over the next century are expected to influence a number of seasonal events including lake ice formation and break-up, snowpack formation, and onset of growing season. Milder winters have resulted in later freeze-up dates, earlier break-up dates, or both for lakes in the northern United States and Canada (Magnuson et al., 2000) and are expected to affect the frequency and severity of ice dams, floods, and low flows in the Arctic. In southwest Alaska, the seasonality of snowpack, lake ice, and vegetation dynamics are expected to have wide- reaching effects. The climate is influenced by complex interactions of several forcing agents, all functioning at different temporal scales: the El Niño and La Niña Southern Oscillation (one-two years), the Pacific Decadal Oscillation (20-30 years) and the Pacific/North American circulation pattern (days to weeks), as well as local forcings resulting from topography (Papineau 2001). These climatic factors influence the timing of freeze-thaw events, lake ice thickness, and surface and groundwater hydrology in the region. Such climate-driven effects on habitat and forage may affect wildlife populations, including caribou, moose, wolves, and a number of other predator and prey species. The presence or absence of lake ice and snow also affects winter access to subsistence activities by rural residents who depend on these resources (e.g., ice fishing, hunting, trapping, firewood collection) (Gaul 2007). Accurate measurements of regional-scale lake ice dynamics, vegetation, and snowpack are needed to improve our understanding of the long-term effects of climate on southwest Alaska ecosystems. The National Park Service’s Inventory and Monitoring (I&M) Program has begun to monitor these variables as a means of characterizing short- and long-term variation in seasonality. Moderate resolution remote sensing data provide a means for monitoring trends in snowpack, lake ice, and vegetation dynamics in this otherwise remote and inaccessible region. The high temporal resolution of MODIS data makes them especially well-suited to capture variation in the timing of seasonal events across years.

Lake Ice Formation and Break-up Process Water density is greatest at 4oC. In the generalized ice formation process, early season autumn snowfall lowers water temperature as snow melts into open water. Surface water cools to 4oC from lowering air temperature and snow. The denser water sinks and the lighter surface water cools until the entire lake mass reaches 4oC. Without further mixing from wind, a lighter layer of water forms on the surface and further cools to 0oC, when a thin layer of skim ice forms (Jeffries et al., 2005). Often, the entire surface of a lake will freeze over within a few hours on a still cold night. The ice layer thickens as additional water freezes to the bottom. Maximum ice thickness varies each winter depending on air temperatures, snow cover and duration of cold weather. Many winters in southwestern Alaska experience mid-winter thaw events as the jet stream moves to the eastern Gulf of Alaska, allowing a strong stream of warm moist air (Chinook storms) to flow north and hit the southern coast of Alaska. This sudden influx of warm air, rain and wind removes the insulating snow cover and the ice surface is flooded with several inches of water. The return of the Arctic high pressure system and cold temperatures refreeze the lake surfaces. This ice cover generally lasts until breakup. Spring breakup of lake ice begins as the days become longer and warmer. Ice decays when the ice becomes isothermal at 0oC and the top and bottom of the ice layer melt simultaneously. Sometimes melting occurs inside the ice layer along the vertical ice crystals, forming columnular or needle ice which rapidly fractures apart, speeding breakup. Thermal absorption from open water, light winds and gentle waves accelerate the melting process. As with freeze-up, breakup can be very rapid, with even large lakes becoming ice free within a few days. Once ice free, lake water gradually warms throughout the summer, reaching 15oC at the surface during warm years. However,

Pecora 17 – The Future of Land Imaging…Going Operational November 18 – 20, 2008 Denver, Colorado water at 50m deep in Lake Clark barely warms to 5oC (NPS unpub data) before the advent of fall storms and the cycle of cooling and ice formation continues.

METHODS

Study Area The study area (Fig. 1) comprises nearly 20 million ha of land between about 56–62° N latitude and 148–160° W longitude in southwest Alaska. It includes the Southwest Alaska Network (SWAN), one of 32 networks in the U.S. National Park Service’s (NPS) Inventory and Monitoring Program (Fancy et al., 2008), which consists of five National Park Service units (Alagnak Wild River, Aniakchak National Monument & Preserve, Katmai National Park & Preserve, Kenai Fjords National Park, Lake Clark National Park & Preserve). This study was initiated as part of the NPS Vital Signs monitoring effort. Southwestern Alaska is characterized by diverse terrain, ranging from gently rolling coastal lowlands to steep, dissected mountain ranges and glaciated valleys. Elevation ranges from 0 to 3100 m. Mean annual precipitation ranges from 400–600 mm in the western interior to 3000–4000 mm at higher elevations and along the eastern coastline of the Cook Inlet Basin (Davey et al., 2007). Mean annual temperature ranges from –8 °C in the western interior to 4 °C in the coastal lowlands, with the growing season generally limited to May–October at lower elevations. Winter temperatures at lower elevations in the region fluctuate around 0 °C. Vegetation in the study area reflects the climatic patterns and is composed primarily of coniferous forests and shrublands along the northern coasts, mixed hardwood-coniferous forests and grasslands at mid-elevations in the interior, and wetlands, tundra, and dwarf shrub communities at higher elevations. Receding glaciers and repeated volcanic activity have resulted in large areas of exposed substrate that are being rapidly colonized by vegetation. For this paper we selected six large lakes to represent lake ice dynamics (freeze-up and break-up) across the range of ecoregions in southwest Alaska (Table 1).

Table 1. Focus areas for vegetation, snow, and lake ice metrics.

Lake Ecoregion Size (hectares) Lake Elevation (m) (Nowacki et al 2002) Skilak Cook Inlet Basin 9,900 60 Tustumena Cook Inlet Basin 29,700 60 Telaquana Lime Hills 4,700 365

Lake Clark Lime Hills 31,000 78 Kukaklek Alaska Peninsula 17,300 300 Nonvianuk Alaska Peninsula 13,200 180

Pecora 17 – The Future of Land Imaging…Going Operational November 18 – 20, 2008 Denver, Colorado

Figure 1. The study area in southwest Alaska with highlighted focus areas.

MODIS Data The MODIS sensors aboard the Terra and Aqua polar-orbiting satellites are part of the National Aeronautics and Space Administration’s (NASA’s) Earth Observing System (EOS). The MODIS sensors acquire data in 36 spectral bands and at three spatial resolutions ranging from 250 to 1000 m (Justice et al., 1997). A swath width of 2330 km enables near-global coverage every day, with multiple swaths collected daily over southwest Alaska. The calibration, spatial detail, and spectral and geolocation quality of MODIS data have improved on its predecessor, the Advanced Very High Resolution Radiometer (AVHRR), which provided 4- to 6-band multispectral data at a resolution of 1.1 km (Townshend and Justice, 2002). A large number of standard, regularly generated products from MODIS are designed for land, ocean, and atmospheric science applications. We made extensive use of the browse images provided by the MODIS Rapid Response System (Rapidfire images), which are created in near-real time for all MODIS granules (http://rapidfire.sci.gsfc.nasa.gov/ ). The natural-color composites of the images were used for the lake ice cover interpretations. We compared interpretations of lake ice cover using Rapidfire images and enhanced composites of the L1 calibrated radiance images. We found that estimates of lake ice metrics were within ± 10 percent cover using the two products, so Rapidfire images formed the interpretation dataset for this research. We also investigated surface reflectance products for lake ice, but a lack of reference aerosol data over our study area resulted in uneven data quality and a large amount of missing data. For the longer term monitoring project, registered 250 m calibrated radiance data will be collected to create a full archive for the study area, and 25 lakes in the region will be monitored for lake ice metrics (Reed et al, 2006). We attempted to use various digital analysis techniques (density slice, several classification approaches) on the georeferenced calibrated radiance data. However, spectral confusion with clouds and fog, topographic and cloud shadows, and ice fracture features like pressure ridges, pans and open leads all contributed to serious misclassification errors. Manual interpretation gave more accurate and faster results when compared with field observations (Table 2). We used density slices of clear-day calibrated radiance data to derive percent ice cover values for several dates with varying ice formation to evaluate the manual interpretations of individual interpreters. Visual interpretations were within ±10 percent of the digitally derived cover values. Percent ice cover was manually interpreted for the six large focus area lakes from daily MODIS Rapidfire natural–color images. These lakes range in size from 4700-31,000 ha (Table 1), far larger than a conservative minimum mapping unit of 100 ha. Seasonal lake ice metrics were calculated from the dates of ice cover: break-up date, freeze-up date and durations of ice season, break-up season, and freeze-up season.

Pecora 17 – The Future of Land Imaging…Going Operational November 18 – 20, 2008 Denver, Colorado Ice appears as white on natural-color images, so snow-covered ice is easily interpreted. Likewise, open water is predominantly black or dark blue on image composites, with lighter blue signifying sediment load from glacial silt or volcanic ash (Fig. 2). Wet ice, which may occur during break-up, by overflow, or from mid-winter rain on solid ice, appears pale blue. Wet ice is often mottled with white and black, reflecting the ice and water mosaic. Another form of blue ice occurs when the frozen-over lake is windswept during winter storms. Pressure cracks are often visible in the imagery, indicating ice rather than open, silty water. Freeze-up date was defined to occur when ice cover was >90 percent, break-up date was when ice cover was <10 percent, and ice season duration was the number of days between freeze-up and break-up (Jeffries et al., in press). There were often small areas (<10 percent) of open water at the lake outlets during freeze-up, or wind stacked ice pans on the windward shore during break-up. These limited areas of residual open water and/or ice were not included in estimates of ice season duration as they are not expected to have much ecological impact on the lake systems.

Figure 2. Spring 2005 break-up of Skilak and Tustumena lakes. The Feb 28 image shows 100 percent ice cover on both lakes. Skilak (upper right corner of image) is predominantly bare ice, while Tustumena is mostly snow covered. By April 1 the ice pack on both lakes is largely fractured into loose pans. By April 17, only remnants of the ice cover persist at the western ends of the lakes.

We analyzed National Climatic Data Center mean daily temperature data from weather stations in the Cook Inlet Basin (Kenai, AK), Lime Hills (Pt. Alsworth, AK), and Alaska Peninsula (King Salmon, AK) to derive the number of days between November 1 and April 30 in each year that had a mean daily temperature of >0 ºC (NCDC, 2008). We also compared graphical depictions of snow and ice depletion curves and growing season phenology to look for synchrony among the metrics (Fig. 3).

Field Verifications Consistent or long term field observations for winter lake ice are scarce in this remote and sparsely traveled portion of Alaska. Sporadic lake ice observations have been recorded for Upper Twin Lake (1962-early 1990’s) by Dick Proenneke (NPS files, Branson 2005), and for Tustumena Lake (approximately 1920-1939) by Andrew Berg (Cassidy and Titus 2003). However, these observations predate satellite data and are most useful to establish trends in ice seasons for earlier decades. Jerry and Jeannette Mills, NPS volunteers live on Telaquana Lake year around and have maintained records for ice cover on the lake from 2001 through the present. Their data are presented in Table 2 as observed ice on and off for Telaquana Lake. Additional observations for various lakes are provided by local residents, pilots and researchers when they are in the country. A temperature array was deployed by the NPS SWAN fresh water monitoring program in the central basin of Lake Clark beginning the winter of 2007-08. The array provides daily temperature and lake level data from near surface to 50m.

Pecora 17 – The Future of Land Imaging…Going Operational November 18 – 20, 2008 Denver, Colorado

Figure 3. Lake ice, snow, and vegetation metrics for the Cook Inlet Basin, Lime Hills, and Alaska Peninsula focus areas. Corresponding daily precipitation totals and mean daily temperatures recorded at Kenai, AK; Pt. Alsworth, AK; and King Salmon, AK are shown for each focus area. The strong El Niño of 2002- 2003 resulted in reduced snow cover and a shortened lake ice season for all focus areas.

Pecora 17 – The Future of Land Imaging…Going Operational November 18 – 20, 2008 Denver, Colorado RESULTS

Within the five lakes reported here, freeze-up dates were more variable (~4–12 weeks) than break-up dates (~2– 5 weeks) (Table 2). Actual start of freeze-up (>10 percent ice cover) was consistent across the region every year. However, start date and duration of full ice cover varied due to dynamics of the freeze-up process in each lake. For example, freeze-up in large, low elevation lakes in the Cook Inlet Basin (e.g., Skilak and Tustumena) was often interrupted by wind or warm weather, but continued during cold calm weather until complete ice cover was formed. Some higher elevation lakes in the Lime Hills and Alaska Peninsula focus areas (e.g., Nonvianuk and Kukaklek) froze early in the season and remained frozen until late in the spring, when they broke up very rapidly. For these lakes, break-up dates were surprisingly consistent across years. Break-up dates on other lakes were much more variable, and in some years the lakes did not freeze at all (e.g., Skilak and Tustumena in 2002–03). Field observations and image interpreted ice on and ice off dates at Telaquana Lake differed by 8-10 days during 2001-03, and 0-3 days for 2004-07 (Table 2). Lake Clark temperature and surface level data (Fig 4) for Oct 2007-May 2008 show water surface temperatures dropping to 0oC on Jan 6-7, with a corresponding leveling of the lake surface. Interpretations of MODIS data for central Lake Clark show the basin open with ice pans on Jan 6, nearly frozen over on Jan 9, and completely ice covered on Jan 10, 2008.

Figure 4. Lake Clark temperature and surface level data for Oct 2007-May 2008 (NPS unpub data).

Pecora 17 – The Future of Land Imaging…Going Operational November 18 – 20, 2008 Denver, Colorado Table 2. Lake ice and snow cover metrics for three focus areas in southwest Alaska. Timing metrics are expressed as calendar date and duration metrics are number of days. A dash (-) indicates years without 100 percent ice cover. The temperature data is expressed as the number of days with mean temperatures greater than 0 oC (d MT > 0 oC).

Neutral Strong Neutral Normal Neutral Weak El Niño El Niño El Niño Cook Inlet Basin Skilak Lake 2001-02 2002-03 2003-04 2004-05 2005-06 2006-07 Ice on – start date 12/7/01 - 1/2/04 1/18/05 1/17/06 1/5/07 Ice off – end date 5/16/02 - 5/3/04 4/25/05 4/30/06 5/8/07 Ice duration 124 0 108 44 99 99 Tustumena Lake 2001-02 2002-03 2003-04 2004-05 2005-06 2006-07 Ice on – start date 12/7/01 - 1/16/04 2/11/05 1/24/06 2/26/07 Ice off – end date 5/16/02 - 5/3/04 5/2/05 4/30/06 5/3/07 Ice duration 146 0 94 33 94 49

Snow on – start date 10/08/01 09/22/02 10/16/03 09/22/04 10/24/05 10/16/06 Snow off – end date 05/09/02 04/15/03 05/01/04 04/23/05 05/09/06 04/23/07 Snow duration (d) 224 216 208 224 208 200

Kenai (d MT > 0oC) 26 80 54 59 50 40 Lime Hills Telaquana Lake 2001-02 2002-03 2003-04 2004-05 2005-06 2006-07 Ice on – start date 11/16/01 12/29/02 12/19/03 12/24/04 11/25/05 11/23/06 Ice off – end date 6/3/02 5/13/03 5/22/04 5/14/05 6/1/06 5/22/07 Ice duration 178 125 153 134 185 161 Observed Ice on 11/25/01 12/29/02 12/24/04 11/25/05 11/23/06 Observed Ice off 5/25/02 5/9/03 5/30/06 5/22/07

Snow on – start date 09/30/01 09/06/02 10/08/03 09/22/04 10/08/05 10/08/06 Snow off – end date 05/25/02 05/25/03 05/17/04 05/09/05 05/25/06 05/01/07 Snow duration (d) 248 272 232 240 240 216

Pt. Alsworth (d MT > 0oC) 54 109 61 81 49 50 Alaska Peninsula Kukaklek Lake 2001-02 2002-03 2003-04 2004-05 2005-06 2006-07 Ice on – start date 12/4/01 12/31/02 12/26/03 12/25/04 12/27/05 12/5/06 Ice off – end date 6/1/02 5/3/03 5/22/04 5/15/05 6/7/06 6/1/07 Ice duration 167 42 146 135 155 169 Nonvianuk Lake 2001-02 2002-03 2003-04 2004-05 2005-06 2006-07 Ice on – start date 12/2/01 12/31/02 12/25/03 12/25/04 11/25/05 11/20/06 Ice off – end date 5/25/02 4/29/03 5/20/04 5/15/05 6/3/06 5/27/07 Ice duration 167 51 128 120 178 181

Snow on – start date 10/08/01 09/30/02 10/08/03 10/24/04 10/08/05 09/30/06 Snow off – end date 05/25/02 05/25/03 06/02/04 06/02/05 06/02/06 06/02/07 Snow duration (d) 240 248 248 232 248 256

King Salmon (d MT > 0oC) 50 98 58 27 35 46

Pecora 17 – The Future of Land Imaging…Going Operational November 18 – 20, 2008 Denver, Colorado DISCUSSION

Preliminary interpretations of the MODIS data suggest that landscape-scale phenomena across the SWAN are complex. Figure 3 shows the time series (2001-2007) of all seasonal metrics that we are recording, along with daily temperature and precipitation, plotted for the three ecoregions. We found that the start of the growing season and the timing of lake ice break-up coincided with snowmelt, but that snowpack formation in the fall did not coincide with the timing of lake ice formation or growing season length. High variability in the timing of freeze-up and break-up, and in the duration of lake ice appeared to be related to variability in winter temperatures suggesting that winter conditions exert a strong influence on all of the metrics measured here. We found a general gradient of variability in the seasonality of the metrics: lake ice was the most variable, followed by snow metrics and vegetation metrics. We found no relationship between the timing of lake freeze-up (ice >90 percent cover) and snowpack formation, although lake ice formation consistently lagged the start of snow season by 2 to 3 months (Fig. 3). The end of snow season and final lake ice break-up dates were generally synchronous. Once mean ambient temperatures rose above 0 °C in the spring, lower elevation lakes in the Cook Inlet Basin focus area (e.g., Skilak and Tustumena) tended to break up within a period of several weeks. Higher elevation, interior lakes in the Lime Hills and Alaska Peninsula focus areas tended to show a greater lag in break-up. Weather records for 2001-2007 show that approximately 15 to 45 percent (Kenai), 30 to 60 percent (Pt. Alsworth), and 20 to 55 percent (King Salmon) of mean daily temperatures between November 1 and April 30 were above 0 ºC (Table 2). El Niño years showed a two- to three-fold increase in the proportion of winter days above 0 ºC, relative to neutral years.

Effect of Winter Conditions on Snow and Lake Ice Season The multiple factors driving the climate in southwest Alaska result in extremely variable temperatures, variation in the form and timing of precipitation, and in the location and strength of winter storms that ultimately control snow season and lake ice through their effect on freezing levels, wind patterns, and periods of freezing temperatures. These storms also affect vegetation metrics through the rate of warming (spring) and cooling (fall). Freeze-up periods were often prolonged as ice on lakes formed during cold weather, was broken apart by the wind or short periods of warm temperatures, then partially froze again, sometimes several times before the lake was completely frozen for the winter. Lake ice cover was generally stable, except during mid-winter (January-February) Chinook storms that partially disrupted the icepack or left several inches of water on top of lake ice. Generally, Arctic high pressure systems return to southwest Alaska in March and April, and the lakes refreeze until final break-up of ice in May and June. In the warming weeks of spring, rising temperatures and longer days lead to the breakdown of the physical structure of ice (Jeffries et al., 2005) and snowpack. Both features become unstable and rapidly disintegrate at around the same time. The period of record reported here has seen alternating El Niño and neutral winters, with a strong El Niño in 2002-03, a normal El Niño in 2004-05 and a weak El Niño in 2006-07 (Papineau, personal communication). Strong El Niños typically result in many warm temperature anomalies in southern Alaska (Papineau 2001). The lakes tended to have shorter duration of ice season during El Niños, with the shortest seasons during the 2002-03 winter, and increasing ice duration in weakening El Niño winters of 2004-05 and 2006-07. Snow season, however, showed longer duration snow cover in the El Niño winters. This may have been caused by heavy precipitation in the fall at higher elevations. The timing and position of late fall storm tracks that bring colder temperatures and snow to the region could be a major contributor to the ice and snowpack characteristics for the entire winter. We expect that relatively small changes in temperature could have regional effects on snowpack and lake ice as winter temperatures fluctuate around 0ºC. The 2002–03 El Niño winter brought temperatures above 0 ºC at elevations below 300 m. The result was rain at lower elevations and a diminished, thin snowpack, while a deep snowpack developed at higher elevations.

Variability in Snow, Lake Ice, and Growing Season Metrics Over the period of this study (2001-2007), lake ice season (freeze-up and break-up dates; ice duration) showed high interannual variability, whereas snow and growing season metrics were more stable across years. In contrast to vegetation metrics, lake ice is highly responsive to short-term weather events, developing overflow and partial ice breakup during windy Chinook storms and freezing overnight when a cold, still high pressure system moves into the area. Snow cover was found to be moderately variable at the beginning of the winter season, whereas duration of snow cover and the end of season (snow off) dates were relatively consistent. Generally, a cold weather system brings snow that remains for weeks to months until warmer air masses move into the region.

Pecora 17 – The Future of Land Imaging…Going Operational November 18 – 20, 2008 Denver, Colorado The timing of these cold weather systems is likely to remain erratic given the speculation for increasing temperature variability. Such variability in thermal regimes could potentially impact the duration of the snow season and the timing of snow cover depletion. The impacts of such changes could be far reaching. Lake ice and snow cover are partial controls on many biological processes, including the growing season metrics reported here, as well as wildlife habitats and movements (Adams et al., 1995; Modafferi and Becker, 1997), salmon migrations and waterfowl nesting (Madsen et al., 2007). For example, the 2002–2003 El Niño winter led to a lack of insulating cover for vegetation at low elevations, wind desiccation, and frost damage. Lack of snow cover limits subnivean travel of microtenes, an important set of prey species, exposing them to increased predation and cold stress (Courtin et al. 1991). Low or no snowpack enables large ungulates such as moose, caribou, and Dall sheep to access a wider range of sparse winter forage, but also allows wolves to travel more efficiently while hunting. Solid ice cover on lakes facilitates human travel for subsistence activities such as ice fishing, hunting, trapping, and wood cutting. Years with poor ice cover yield limited harvests and, often, accidents involving travelers going through the ice.

SUMMARY

A set of ecologically oriented metrics were derived from a series of MODIS products to assess interannual variability in snowpack, lake ice, and vegetation dynamics in southwest Alaska. The lake ice and snow metrics reported here appeared to be strongly influenced by global and regional climate and local topography, and represent the dynamics of the physical processes on the SWAN landscape. Lake ice showed the greatest temporal variation of the observed seasonal metrics, followed by snow cover and then vegetation metrics. This pattern is likely due to the factors controlling ice, snow and vegetation. The derivation and analysis of separate but related sets of satellite–generated, ecologically meaningful metrics gives us the opportunity to examine ecosystem function at a scale that has not been previously possible. For long-term monitoring, we need simple analysis techniques and easily interpreted metrics that are neither sensor nor format specific. However, we do need sensor packages that are of moderate spatial resolution and high temporal resolution with visible, near-infrared, and mid-infrared wavelengths. Continuity of such sensor systems is critical to long-term ecological monitoring. The data provide an overview of ecosystem-wide seasonal dynamics that can be compared across years. The length of record currently imposes limitations on the types of conclusions that can be drawn, but the growing data archive has already provided evidence of potential linkages among snowpack, lake ice, and vegetation dynamics.

ACKNOWLEDGEMENTS

We acknowledge the contributions of Jeanette and Jerry Mills, National Park Service volunteers-in-residence at Telaquana Lake, for their meticulous records of lake ice cover and associated weather records for the last 8 years. Their data have served as a gratifying confirmation of the ice season interpretations and provided explanations for unusual ice phenomena. We appreciate John Papineau, NOAA, for patient discussions about climate patterns in SW Alaska. Jeff Shearer, SWAN aquatic ecologist provided the temperature array data fro Lake Clark and collaborative discussions about freeze-up. This project was funded by the National Park Service Inventory and Monitoring Program, Southwest Alaska Network .

REFERENCES

Adams, L.G., Singer, F.J., and Dale, B.W., 1995. Caribou calf mortality in Denali National Park, Alaska, The Journal of Wildlife Management, 59, 584-594. Branson, J. (ed)., 2005. More Readings from One Man’s Wilderness. The Journals of Richard L. Proenneke 1974- 1980, USDOI, NPS, Lake Clark NP&P, AR/CCR-2005-53, Anchorage, AK. Cassidy, C. and G. Titus, 2003. Alaska’s No.1 Guide, The History and Journals of Andrew Berg, 1886-1939, Spruce Tree Publishing, Soldotna, AK. Courtin, G.M., Kalliomaki, N.M., Hillis, T., and Robitaille, R.L., 1991. The effect of abiotic factors on the overwintering success in the meadow vole, Microtus pennsylvanicus: Winter redefined, Arctic and Alpine

Pecora 17 – The Future of Land Imaging…Going Operational November 18 – 20, 2008 Denver, Colorado Research 23, 45-52. Davey, C.A., Redmond, K.T., and Simeral, D.B., 2007. Weather and Climate Inventory, National Park Service, Southwest Alaska Network, Natural Resource Technical Report NPS/SWAN/NRTR—2007/045, 80 pgs. Fancy, S. G., J. E. Gross, and S. L. Carter, 2008. Monitoring the condition of natural resources in U. S. National Parks, Environmental Monitoring and Assessment, in press. Gaul, K.K., 2007. NANUTSET ch’u Q’udi Gu, Before Our Time and Now, An Ethnohistory of Lake Clark National Park and Preserve, USDOI, NPS, Lake Clark National Park and Preserve, 179 pgs. Jeffries, M.O., Morris, K., and Kozlenko, N., 2005. Ice characteristics and processes, and remote sensing of frozen rivers and lakes, In C.R. Duguay and A. Pietroniero (Eds.), Remote Sensing in Northern Hydrology: Measuring Environmental Change (pp. 63-90), Washington, DC, American Geophysical Union, AGU Monograph. Justice, C.O., Vermote, E., Townshend, J.R.G., Defries, R., Roy, D.P., Hall, D.K., Salomonson,V.V., Privette, J.L., Riggs, G., Strahler, A., Lucht, W., Myneni, R.B., Knyazikin, Y., Running, S.W., Nemani, R.R., Wan, Z.M., Heute, A.R., van Leeuwen, W., Wolfe, R.E., Giglio, L., Muller, J.P., Lewis, P., and Barnsley, M.J., 1997. The moderate resolution imaging spectroradiometer (MODIS): Land remote sensing for global change research, IEEE Transactions on Geoscience and Remote Sensing, 36, 1228-1249. Madsen, J., Tamstorf, M., Klaassen, M., Eide, N., Glahder, C., Rigét, F., Nyegaard, H., and Cottaar, F., 2007. Effects of snow cover on the timing and success of reproduction in high-Arctic pink-footed geese Anser brachyrhynchus, Polar Biology, 30 (11), 1363 - 1372. Magnuson, J.J., Robertson, D.M., Benson, B.J., Wynne, R.H., Livingstone, D.M., Arai, T., Assel, R.A., Barry, R.G., Card, V., Kuusisto, E., Granin, N.G., Prowse, T.D., Stewart, K.M., and Vuglinski, V.S., 2000. Historical trends in lake and river ice cover in the northern hemisphere, Science, 289, 1743-1746. Modafferi, R.D., and Becker, E.F., 1997. Survival of radiocollared adult moose in Lower Susitna River Valley, southcentral Alaska, The Journal of Wildlife Management, 61, 540-549. Nowacki, G., Spencer, P., Fleming, M., Brock, T., and Jorgenson, T., 2002. Unified Ecoregions of Alaska B2001, Digital data available: http://agdc.usgs.gov/data/projects/fhm/USGS Open File Report 02-297. 1 map. NPS. Unpub data. Lake Clark temperature array data, SWAN files, Anchorage, AK. NPS files. Proenneke Collection, Original Annotated Calendars, Lake Clark and Katmai Study Center, Anchorage, AK. Papineau, J.M., 2001. Wintertime Temperature Anomalies in Alaska Correlated with ENSO and PFO, Int. J. Climatol, 21: 1577-1592. Reed, B.C, M. Budde, P. Spencer, and A. Miller, 2006. Satellite-Derived Measures of Landscape Processes: Monitoring Protocol for the Southwest Alaska I&M Network, National Park Service, Inventory & Monitoring Program, Southwest Alaska Network, Anchorage, Alaska, 30 pp. Reed, B.C, M. Budde, P. Spencer, and A. Miller. Integration of MODIS-derived metrics to assess interannual variability in snowpack, lake ice, and NDVI in southwest Alaska, Remote Sensing of the Environment, in press. Townshend, J.R.G. and C.O. Justice, 2002. Towards operational monitoring of terrestrial systems by moderate resolution remote sensing, Remote Sensing of Environment, 83, 351-359.

Pecora 17 – The Future of Land Imaging…Going Operational November 18 – 20, 2008 Denver, Colorado